About me

My name is Lin Gong and I am a final year graduate student in the Department of Computer Science, Univeristy of Virginia, under the guidence of Dr.Hongning Wang. My research interest lies in data mining and machine learning, especially in sentiment analysis and social influence analysis. I would like to explore how user-generated data such as textual reviews and social connections help better understand their intents.

Contact Info.

Email: lg5bt@virginia.edu
Website: www.cs.virginia.edu/~lg5bt/
Address: Rice Hall 224, Charlottesville, VA 22903


Aug, 2018

I participated KDD 2018 in London, United Kindom.

May, 2018

Our paper got accepted by KDD 2018.

May, 2017

We are the winners of Yelp Data Challenge Round Eight!

Dec, 2016

Our paper got accepted by WWW 2017.

May, 2016

Our paper got accepted by ACL 2016.

April, 2016

Our proposal "Predicting Community-Level Criminal Behaviors by Estimating Human Attitudes from Social Media" was selected for the Presidential Fellowship in Data Science!

April, 2016

I participated the CRA-W 2016, San Diego and presented my work in the poster section.

Oct, 2015

I participated the Grace Hopper 2015 in Houston, Texas.


Web Design Trends

Holistic User Behavior Modeling

Guided by Hongning Wang    Jan, 2017 - Jan, 2018

We focus on self-consistence across multiple modalities of user-generated data to model user intents. A probabilistic generative model is developed to integrate two companion learning tasks of opinionated content modeling and social network structure modeling for users. Individual users are modeled as a mixture over the instances of paired learning tasks to realize their behavior heterogeneity, and the tasks are clustered by sharing a global prior distribution to capture the homogeneity among users.

Web Design Trends

Clustered Model Adaptation for Personalized Sentiments

Guided by Hongning Wang    July, 2016 - Dec, 2016

We propose to capture humans’ variable and idiosyncratic sentiment via building personalized sentiment classification models at a group level. Our solution roots in the social comparison theory and the cognitive consistency theory. We formalize personalized sentiment classification as a multi-task learning problem. In particular, to exploit the clustering property of users’ opinions, we impose a non-parametric Dirichlet Process prior over the personalized models, in which group members share the same customized sentiment model adapted from a global classifier.

Web Design Trends

Multi-task Model Adaptation for Personalized Sentiment Classification

Guided by Hongning Wang    Jan, 2016 - June, 2016

Motivated by the findings in social science that people’s opinions are diverse and variable while together they are shaped by evolving social norms, we perform personalized sentiment classification via shared model adaptation over time. In our proposed solution, a global sentiment model is constantly updated to capture the homogeneity in which users express opinions, while personalized models are simultaneously adapted from the global model to recognize the heterogeneity of opinions from individuals.

Collaborative Model Adaptation for Personalized Opinion Analysis

Guided by Hongning Wang    June, 2015 - Dec, 2015

Humans' opinions are variable: the same opinions can be expressed in various ways and the same expression can carry distinct sentimental polarities. Thus, a global sentiment classification model is incompetent to distinguish such diverse opinions while personalized models are limited to sparsity . we propose to build personalized sentiment classification models via adapting a global model to individual users in a collaborative manner. The learned sentiment models are shared across neighboring users to conquer data sparsity challenge.

Network-based Sentiment Analysis for Online Reviews

Guided by Hongning Wang    Nov, 2014 - May, 2015

Supervised learning methods suffer in performance with limited amount of labeled data. However, labeling data is quite expensive and requires expertise. By constructing network and sharing information among reviews, we can better propagate the sentiment in reviews with limited amount of labeled data. We utilized learning to rank technique to further purify the neighborhood by extracting more representative features from reviews, i.e., POS tags, aspect vector, topic distribution. Then, we adopted transductive learning method to propagate the sentiment labels among documents to boost the performance.

Analysis of Amazon User Reviews and Prices.

Cooperated with Lingjie Zhang, Bo Man    Sep, 2014-Dec, 2014

User reviews are gaining more and more attention. Retailors need reveiws to get feedback while users need reviews to acquire more information. On one side, retailors may perform comerical campaigns based on user's feedback, i.e., sales. On the other side, users' phurchase decision may reply on others' reviews laregely with more data available. Thus, we were curious are there any relationships between user reviews and prices. Read our slides to get a quick view of what we have done and more deatails are described in our report.


[1] Lin Gong and Hongning Wang. When Sentiment Analysis Meets Social Network: A Holistic User Behavior Modeling in Opinionated Data. The 24th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD 2018), p1455-1464, 2018.[pdf][git][data][video]

[2] Lin Gong, Benjamin Haines and Hongning Wang. Clustered Model Adaptation for Personalized Sentiment Analysis. The 26th International World Wide Web Conference (WWW 2017), p937-946, 2017. [pdf][code][documentation]

[3] Lin Gong, Mohammad Al Boni and Hongning Wang. Modeling Social Norms Evolution for Personalized Sentiment Classification. The 54th Annual Meeting of the Association for Computational Linguistics (ACL 2016), p855-865, 2016. [pdf][code][documentation]


The modeling method of parallel power amplifier based on improved BP neural network. (China, No. 201210443798.8)